This track no longer gives "poor people" a chance: If you don't have a billion, don't talk about Series B

Guide Read THECAPITAL Fire and oil, surrounded by flowers This article is 4643 words long, approximately 6.6 minutes Author | Wang Tao Editor | Us Source | #Rongzhong Finance (ID: thecapital) In just two weeks after the Spring Festival Gala, robots are once again making headlines, this time not for their stage performances, but for the overwhelming influx of financing information. According to incomplete statistics from Rongzhong Finance reporters, the humanoid robot industry (including components) has completed at least 18 financing rounds, with a total amount exceeding 13 billion yuan; data from IT Juzi shows that since the beginning of 2026, there have been a total of 88 disclosed financing events in the field of embodied intelligence in China, with a total financing amount exceeding 20 billion yuan. What is even more astonishing is that in recent months, the financing situation for robots has seen Series A financing reach the 1 billion yuan threshold, with subsequent Series B financing amounts mostly at or above the 1 billion yuan level. Generally speaking, a scale of 1 billion yuan mainly appears in Series C or even later rounds, but this time it has become the basic threshold for star robots. (Rongzhong Finance Graphics) This inevitably raises the question: Is the humanoid robot track truly promising, or is capital blindly following trends? In the reporter's view, this 1 billion yuan threshold is a line drawn by capital with real money; crossing this line means that the company may have the conditions for a commercial closed loop and has a ticket to the next stage. The 1 Billion Yuan Threshold In 2026, the Chinese robot track is no longer a "small-scale" entrepreneurial incubator. By the end of 2025, Yuanli Lingji's Series A financing surpassed 1 billion yuan, setting a new height for Series A financing and becoming the first domestic embodied intelligence company centered around models. In just over two months into the new year, the market has been inundated with news of large financing rounds. On February 2, LimX Dynamics announced the completion of a $200 million (approximately 1.44 billion yuan) Series B financing. This round attracted e-commerce and cloud computing giants, top global hard tech VCs, leading automotive manufacturing industry chains, and several local government industrial guidance funds, including Alibaba, Sequoia China, SAIC Investment, Lenovo Capital, and Shenzhen Capital Group. On February 23, AI²Robotics announced the completion of over 1 billion yuan in Series B financing. The lineup of investors in this round spans internet and AI giants, leading state-owned enterprises, leaders in the Tesla ecosystem, strategic partners in various scenarios, "deep-pocketed" PE funds, leading brokerage firms, and local funds, including Baidu, CRRC Capital, Yuxin Technology, Senqilin, Yunbo Capital, and Guotai Junan, making for a very luxurious lineup On March 2, Galbot announced the completion of a 2.5 billion yuan Series B financing. The lineup of investors in this round is extremely strong, making it a "national-level" benchmark in the field of embodied intelligence, including national industrial investment funds, energy and petrochemical giants, top state-owned commercial banks, and local strategic investment platforms, such as the National Artificial Intelligence Industry Investment Fund (National Big Fund Phase III), Sinopec, Bank of China, Industrial and Commercial Bank of China, and Beijing Artificial Intelligence Industry Guidance Fund. In addition, a number of players such as Xinghaitu, Lingxin Qiaoshou, and Zhujidongli have all crossed the 1 billion yuan mark. It can be seen that in the field of embodied robotics, 1 billion yuan has basically become the "entry threshold" for Series B financing. 2025 is referred to as the year of mass production for robots, and at that time, robot financing can be described as overwhelming. In the first two months of this year alone, the total amount has exceeded 13 billion yuan, approaching one-third of the total for the entire year of 2025. This shows that, with the boost from the Spring Festival Gala, humanoid robots are firing on all cylinders this year. From the perspective of investors, industrial capital and state-owned capital have become the main forces in this wave of financing. In summary, Meituan has invested in Galbot, Zhujidongli, Zivariable Robotics, Qianxun Intelligent, Zhijian Dongli, and Xingdong Jiyuan, almost covering all leading companies in the field, aiming to secure all possibilities for future unmanned delivery and smart warehousing. ByteDance has invested in Zivariable Robotics (leading) and Galbot. Logically, the focus is primarily on algorithms, with fewer but highly precise investments directed towards teams that possess end-to-end large model algorithms and can directly generate synergies with Byte's AI laboratory. Additionally, Songyan Dongli has backing from CATL, Zhujidongli is supported by JD.com and SAIC, while Baidu's strategic investment and CRRC Capital also appear on the shareholder list of Zhifang Square, which are typical cases of deep participation by industrial capital. The presence of state-owned capital is also becoming increasingly active, with the Jingguosheng Fund behind Songyan Dongli, Chengdu Science City Venture Capital and Kesheng Fund behind Zhifang Square, and the National Artificial Intelligence Industry Investment Fund (Big Fund Phase III) backing Galbot. When the "national team," which holds national strategic resources and industrial lifelines, begins to bet real money, it signifies that the humanoid robot sector is no longer a "bubble" pursued by capital but is becoming a tangible cornerstone in the transformation and upgrading of China's manufacturing industry. Perhaps this is the confidence that many investors have in placing their bets. As for the use of funds from the Series B round, most of these robotics companies seem to relate their answers to industrial implementation. For example, Zhifang Square stated that the funds from this financing will mainly be used to maintain the leading capability of the GOVLA embodied large model and, based on this, drive the iteration and production line expansion of the AlphaBot series of robotic products. Lingxin Qiaoshou also stated that the funds raised in Series B will be invested in core product technology research and development, capacity ramp-up, and full-stack base capability construction. Some believe that 2026 will be the year of large-scale volume for embodied robots "from laboratory to factory." To truly achieve large-scale implementation, funding is a very important prerequisite. Therefore, capital is willing to place heavy bets in Series B. 1 billion yuan is a watershed for capital; crossing it means that the enterprise has the potential to enter the final circle of robots on a large scale A billion in Series B is not just a round of financing, but a ticket to the next stop. Behind the "Hundred Billion Game," Models Become the Competitive Advantage As the financing threshold in the embodied intelligence industry generally crosses the billion level and valuations approach the hundred billion mark, the aesthetic logic of capital has undergone a fundamental shift. In the years-long technological marathon, securing funding is merely an "entry ticket," while the real moat is no longer simply the depth of algorithms or hardware parameters, but the general intelligence of the "brain," architectural infrastructure, and extremely pragmatic engineering implementation capabilities. In the early robotics track, capital was often attracted by precise mechanical structures, high-torque joints, or flexible dexterous hands. However, as the supply chain matures, core components such as motors, reducers, and actuators are gradually moving towards modularization and standardization, and the differentiated advantages of hardware are rapidly leveling out. Previously, Yuanli Lingji stated in an interview with RZhong Finance: "Currently, robots face setbacks in real environments, and the core shortcoming is not the torque of the motors or insufficient degrees of freedom of the hardware, but rather 'insufficient IQ' (lack of physical intuition and poor generalization)." It is reported that Yuanli Lingji is an embodied intelligence company established in March 2025, focusing on the research and development and industrialization of full-stack robotics hardware and software technology. They proposed the "embodied native" technology route and released core products such as the embodied native large model DM0, the open-source development framework Dexbotic, and the mass production workflow DFOL. The company has completed nearly 1 billion yuan in financing led by Alibaba, Nio Capital, and others. The "insufficient IQ" mentioned by Yuanli Lingji manifests as robots lacking physical intuition and generalization ability. A robot that can walk steadily on a clean laboratory floor often stops working when it enters a real factory with variable lighting and clutter due to perception failure or decision-making collapse. Therefore, in 2026, capital reached a strong consensus: the true moat at this stage is "the general intelligence of the brain and architectural infrastructure." Yuanli Lingji stated: the differentiated advantages of hardware will gradually level out, and only embodied native models with cross-model generalization capabilities can be downward compatible with various hardware and upward adaptable to myriad scenarios. This path of "model unlocking scenarios, scenarios defining hardware" has made "first the model, then the entity" the core trump card of leading enterprises. In the hundred billion-level game, investors have become unprecedentedly calm. They no longer pay for exquisite demo videos but instead use a magnifying glass to search for companies with the following three "hard capabilities": Full-stack self-research and mass production delivery capability: Key components such as joints, dexterous hands, actuators, and chassis must possess full-stack self-research capabilities, which is the foundation for ensuring that the supply chain does not "choke." However, self-research is just the first step; more importantly, it is the engineering system capability. "Reasoning must be sufficiently real-time, the whole machine must be stable and reliable, and the safety mechanisms must be complete; at the same time, the 'dirty and tiring work' of cost, loss, maintenance, and deployment must be done well, otherwise, no matter how strong the algorithm is, it cannot be implemented. Only companies that can stably go offline, have low failure rates, and can deliver at scale meet the first threshold for crossing cycles." Embodied large models and real data closed loop: Capital is no longer focused solely on algorithms, but on "intelligence of the real world" plus "data closed loop." "This intelligence means that robots must not only complete tasks in controlled environments but also succeed stably in different objects and environments, without task sequences collapsing, and be able to interact naturally with humans. More critically, companies must be able to feed both good and bad examples from the field back into training, making the model more accurate with use. This self-evolution capability is the key to continuously widening the gap." Real orders and commercial return on investment: 2026 is the "ROI (Return on Investment) year" for robot commercialization. Capital will only invest in companies that "can make money," no longer paying for concepts. Companies must prove they can truly solve a scenario's problems from start to finish, calculate ROI, and ensure that delivery methods are standardized, replicable, and scalable. Missing any one of these aspects makes it difficult to succeed in this long race. This wave of financing starting at one billion and valuing at hundreds of billions is essentially not a bubble, but a qualitative change point in the industry moving from 0 to 1 to 1 to 10. The entry of national teams and the clustering of industrial capital mean that embodied intelligence has transformed from a "high-end toy" in the hands of geeks to a "productive weapon" in great power competition. At this turning point, those who survive will certainly be the players with the most extreme "soft and hard integration." The companies that will emerge in the future will be those that can manufacture hardware, have strong brains, and run commercially. Hardware ensures the "lower limit" of robots, determining whether they can enter the market; while the brain determines the "upper limit" of robots, deciding how much work they can do and how intelligently they can perform. This ultimate alignment of "brain" and "body" is not only a technical challenge but also a comprehensive test of team organizational capability, capital utilization efficiency, and industry insight. Capital is clearing the field in advance through extremely high financial thresholds, rapidly concentrating resources on leading projects with this composite capability. For players involved, securing one billion in financing is just the first step to survival; how to convert this large sum into stable performance in real scenarios and a continuous stream of orders is the true "trump card" that determines life and death. From assembly line to "Android moment" After crossing the capital threshold and accumulating underlying strength, the embodied intelligence industry faces the most realistic question: Where is the first foothold for large-scale commercialization? The debate over whether to "enter factories first" or "enter homes first" is gradually being clarified by market feedback in 2026. Original Force Intelligence indicates that the first commercial ROI (Return on Investment) explosion point will occur in B-end factories and specific industrial scenarios. This judgment not only reveals the objective laws of technology landing but also predicts the inevitable trend of embodied intelligence evolving from a "single tool" to a "platform ecosystem." The scientific assessment standard for scenario landing lies in the fault tolerance rate and the degree of data structuring. Home environments are seen as extremely unstructured "open worlds," filled with a vast number of long-tail scenarios (Corner Cases). "There are many Corner Cases in home environments, and C-end users have a very low tolerance for failure, making it difficult for large models to support 100% generalization in the short term." In a home setting, a broken cup or a wrong item retrieval instruction can lead to a collapse of user confidence. This high brand risk, combined with the difficulty of technology generalization, means that "home entry" is still largely in an exploratory phase at this stage. In contrast, factory assembly lines represent a semi-structured environment. Here, tasks are highly repetitive, boundaries are clear, and production rhythms are quantifiable. Enterprise clients only pay for "efficiency and cost reduction," making ROI easy to calculate. In industrial scenarios, even if robots occasionally make mistakes, the controlled environment and comprehensive safety mechanisms ensure that the resulting losses are manageable and predictable. More importantly, by honing "accuracy and execution rhythm" in B-end factories at a high frequency, companies can run through the core data flywheel, which is seen as a pragmatic foundation for embodied intelligence to eventually reach C-end households. After securing substantial financing, players have also shown significant differentiation in their role settings. Some companies choose to build vertical integrated robot benchmarks, while others look further ahead. For example, Yuanli Lingji does not pursue being a closed "vertical hardware manufacturer"; its core positioning is as a provider of embodied brains and an open-source builder of infrastructure in the era of embodied intelligence. This shift in positioning essentially attempts to upgrade from a business of "selling a machine" to "defining an era." This transformation is vividly compared to the "Android moment" or "PyTorch moment" in the field of embodied intelligence. Just as the popularity of smartphones relies on the open-source Android system, the large-scale explosion of embodied intelligence also requires a standardized "architectural infrastructure." By releasing an embodied native development framework, establishing a global standard real machine evaluation platform, and open-sourcing foundational models and dual-arm hardware, leading companies are striving to break down industry silos. Yuanli Lingji states that through these open-source and standardization efforts, it hopes to empower developers and hardware manufacturers across various industries, aiming to become a cornerstone ecosystem builder that drives AI intelligence to truly take root in the physical world. In 2026, this industry inflection point, the capital frenzy is not creating a bubble but accelerating the industry's leap from 1 to 10. Companies that can truly last in this marathon must possess three hard capabilities. First is "intelligence in the real world" plus "data closed loop," requiring models not only to run demos but also to succeed stably in different environments and to return on-site data for real-time training. Second is strong engineering capability, able to handle maintenance, deployment, and other "dirty and tiring work," ensuring that algorithms are truly implemented. Finally, there is product and commercial capability, ensuring that delivery methods can be scaled. In summary, the billion-dollar financing threshold is merely a ticket to entry, while the real competition lies in who can first run through ROI in B-end factories and who can first build the underlying infrastructure for embodied intelligence. This is not only a technological game but also a contest of business vision and ecological appeal. In the coming years, those who survive and lead the industry will undoubtedly be the companies that can perfectly align the generalization capabilities of the "brain" with the engineering quality of the "body" and build a foundational ecosystem with an open mindset. The curtain has just been raised, and the finale will belong to those who truly root AI in the physical world through practical efforts Clue Disclosure # rzcj@thecapital.com.cn Media Cooperation: 010-84464881